JP2007049753A - Single-user detection - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To simplify user equipment in multiplex communication by shared frequency spectrums of a TDD/CDMA communication scheme or the like. <P>SOLUTION: In a CDMA communication system, a plurality of data signals are transmitted from the transmitting side via the shared frequency spectrums. The transmitted data signals receive similar channel responses. On the receiving side, the combined signal of the transmitted data signals is received. The composite signal is sampled at a speed which is a multiple of the chip velocity. A channel response to the composite signal is calculated. A composite signal sample value and a channel response estimate are used to calculate the first element of a spread data vector. The coefficient from the calculation of the first element is used to calculate the remaining elements of the spread data vector. The calculated elements of the spread data vector are used to calculate the data of the data signals. <P>COPYRIGHT: (C)2007,JPO&INPIT

Description

  The present invention generally relates to wireless communication systems. More particularly, the present invention relates to data detection in a wireless communication system.

FIG. 1 shows a wireless communication system 10. The wireless communication system 10 includes base stations 12 _{1 to} 12 ₅ (12) that communicate with user equipments (UEs) 14 _{1 to} 14 ₃ (14). Each base station has a corresponding operating range within which the base station communicates with the user equipment UEs14.

  In some communication systems, for example, time division duplex (TDD / CDMA) communication systems using code division multiple access (CDMA) or code division multiple access, multiplex communication is performed with the same frequency spectrum. These multiplex communication signals are distinguished by using different codes for each channel. In order to use the radio frequency spectrum more efficiently, the TDD / CDMA communication system uses repetitive frames divided into a plurality of time slots for communication. Each communication signal sent in such a system has one or more corresponding codes and a time slot assigned to it. One code use in one time slot is called a resource unit.

IEEE International Symposium on Personal, Indoor and Mobile Radio Communications XX, DD vol.3, 1998, pages 1340 to 1345 IEEE Transactions on Information Theory, March 1994, vol.40, No.2, 340 to 349 International Annual Conference of ICT, XX, XX, 1999, 1-7

Since multiplex communication is performed simultaneously in the same radio frequency spectrum, a receiver in such a communication system needs to distinguish a large number of communication signals from each other. One technique for detecting such signals is the multi-user detection method. In the multi-user detection method, signals corresponding to all users, that is, user apparatuses UEs 14 are detected simultaneously. Techniques for performing multi-user detection include block linear equalization (BLE-JD) based on simultaneous detection using Cholesky decomposition or approximate Cholesky decomposition. These detection means are associated with a high degree of complexity. This complexity increases the power consumption and thus reduces the battery life of the user equipment UEs14. Therefore, development of alternative means for detecting received data is required.

  In a code division multiple access communication system, the transmission side transmits a plurality of data signals via a shared frequency spectrum. Each transmitted data signal undergoes a similar channel response. On the receiving side, a composite signal of the transmitted data signals is received. The synthesized signal is sampled at a multiple of the chip speed. Thereby, a common channel response to the received composite signal is estimated. Next, the first element of the spread data vector is calculated using the sample value of the composite signal and the estimated channel response value. The remaining vector elements of the diffusion data vector are also calculated using the coefficients from the calculation of the first vector element. Data of the data signal is calculated using the calculated diffusion data vector element.

  This simplifies the CDMA communication receiver using the CDMA system and the shared frequency spectrum, and enables downsizing the receiver and reducing power consumption.

FIG. 2 shows in simplified form a transmitter 26 and a receiver 28 using a single user detection (SUD) technique in a TDD / CDMA communication system. The single user detection method can also be applied to communication systems such as other communication systems and frequency division duplex (FDD) CDMA. In a normal communication system, the transmitter 26 is in each user apparatus UE14, and the multiplex transmission circuit 26 that performs multiplex communication is in each base station 12. The SUD receiver 28 can be located at the base station 12, the plurality of UEs 14, or both. Usually, the SUD technique is used to detect data during single-code or multi-code transmission from a specific transmitter. If all transmitted signals are sent from the same transmitter, each individual channel code signal will receive the same channel impulse response in multi-code transmission. The SUD technique is particularly useful for the downlink, that is, the downlink in which all transmission signals are radiated from one antenna or antenna array of one base station. This approach is also useful for the uplink where a single user occupies one slot for single code transmission or multi-code transmission.

  The transmitter 26 sends data via the wireless transmission channel 30. A data generator 32 in the receiver 26 produces data to be transmitted to the receiver 28. The modulation / spreading sequence insertion unit 34 spreads the data and time-division-multiplexes the spread reference data with a midamble training sequence and a data spreading code within an appropriate allocation time slot, thereby providing one or a plurality of communication signals. Generate a burst.

  As shown in FIG. 3, the normal communication signal burst 16 includes one midamble 20, one guard period 18, and two data areas 22 and 24. The midamble 20 separates the two data areas, and the guard period 18 separates the communication signal bursts to allow for differences in burst arrival times from different transmitters 26. The two data areas 22 and 24 contain communication signal burst data.

  The modulator 36 modulates the radio frequency (RF) with these communication signal bursts. The RF signal is radiated by the antenna 38 and reaches the antenna of the receiver 28 via the wireless transmission channel 30. The modulation method used for transmission of the communication signal may be any method known to those skilled in the art, such as quadrature phase shift keying (QPSK) and Mth order quadrature phase modulation (QAM).

  The antenna 40 of the receiver 28 receives various RF signals. The received RF signal is demodulated by the demodulator 42 to reproduce the baseband signal. The baseband signal is sampled at a chip rate of the transmission burst signal or at a multiple of the chip rate by a sampling device 43 such as a single type or composite type A-D converter. The sampled output is processed by a device such as channel estimator 44 or SUD device 46 with the appropriate code within that time slot assigned to the received signal burst. The channel estimator 44 uses the midamble training sequence component in the baseband samples and provides channel information such as channel impulse response. This channel impulse response is represented as matrix H. This channel information is used by the SUD device 46 to estimate the transmitted data in the soft symbol received communication burst.

The SUD device 46 uses the channel information from the channel estimation device 44 and the known spreading code used by the transmitter 26 to estimate the desired received communication burst data. Here, the SUD will be described using the Universal Terrestrial Radio Access (UTRA) TDD system of the 3rd Generation Partnership Project (3GPP) as a background communication system, but this SUD can also be applied to other systems. The system is a direct spread wideband CDMA (W-CDMA) system, that is, a wideband CDMA (W-CDMA) in which an uplink signal and a downlink transmission signal are limited to separate time slots.
The receiver 28 receives a total of K burst signals using the antenna 40 (48). These K burst signals are superimposed on each other within one observation interval. For the 3GPP UTRA TDD system, each data region of a time slot corresponds to one observation interval.

  For one observation interval, the data detection task is represented by Equation 1.

ここで、ｒは受信したサンプル値で、Ｈはチャネル応答行列、ｄは拡散データベクトルである。拡散データ行列は、各チャネルで伝送されてきたデータとそのチャネルの拡散符号とを含んでいる。

Here, r is a received sample value, H is a channel response matrix, and d is a spread data vector. The spread data matrix includes data transmitted through each channel and the spread code of that channel.

If the received signal is oversampled, a plurality of sample values for each transmission chip are generated, resulting in reception vectors r ₁ ,..., R _N (48). Similarly, the channel estimation device 44, the received vector r _1, ..., the channel response H ₁ corresponding to r _N, ..., to calculate the H _N (50). For sampling at twice the chip speed, Equation 1 is rewritten as Equation 2.

ここで、ｒ₁は偶数番目のサンプル(チップ速度でのサンプリングに対応)で、ｒ_２は奇数番目のサンプル(ｒ₁サンプルからチップ間隔の二分の一離れた所でのサンプリングに対応)である。Ｈ₁は奇数番目のサンプルに対するチャネル応答で、Ｈ_２は偶数番目に対するその応答である。

Here, r ₁ is an even-numbered sample (corresponding to sampling at the chip speed), and r ₂ is an odd-numbered sample (corresponding to sampling at a half distance of the chip interval from r ₁ sample). . H ₁ is the channel response for odd-numbered samples and H ₂ is the response for even-numbered samples.

  Equation 1 becomes Equation 3 when sampling at a speed N times the chip speed.

ここで、ｒ₁、ｒ₂,…、ｒ_Nは、チップ速度のそれぞれの倍数でのサンプルである。各サンプルはチップ間隔のＮ分の１だけずれている。Ｈ₁、Ｈ₂、…、Ｈ_Nは対応するチャネル応答である。以下の説明はチップ速度の２倍の速度でのサンプリングを実行する受信機に絞るが、同様な手法はチップ速度の任意の倍数でのサンプリングに対しても適用できる。

Here, r ₁ , r ₂ ,..., R _N are samples at respective multiples of the chip speed. Each sample is offset by 1 / N of the chip spacing. H ₁ , H ₂ ,..., H _N are the corresponding channel responses. The following description focuses on a receiver that performs sampling at twice the chip rate, but a similar approach can be applied to sampling at any multiple of the chip rate.

For sampling at twice the chip rate, the magnitudes of the channel response matrices H ₁ and H ₂ are (Ns + W−1) × Ns. Here, Ns is the number of spreading chips transmitted within the observation period, and W is the length of the channel impulse response, for example, 57 chips in length. Since the received signal has the Ns diffusion tip, the length of r ₁ and r ₂ are Ns. Equation 2 can be rewritten as Equation 4.

ｒ₁(i)、ｒ₂(i)、ｈ₁(i)及びｈ₂(i)は、それぞれ対応するベクトル行列ｒ₁、ｒ₂、Ｈ₁及び
Ｈ₂のｉ番目の要素である。
拡散データベクトルを算定する手法の一つは、拡張前進代入手法であり、その手法を図４を参照して説明する。拡張前進代入のために、受信したデータベクトルを再配列し、各偶数番目のサンプルの次にそれに対応する奇数番目のサンプルを伴うようにする。同様の再配列を式５aに示すとおりチャネル応答行列についても行う。

r ₁ (i), r ₂ (i), h ₁ (i) and h ₂ (i) are the i-th elements of the corresponding vector matrices r ₁ , r ₂ , H ₁ and H ₂ , respectively.
One of the methods for calculating the spread data vector is an extended forward substitution method, which will be described with reference to FIG. For extended forward substitution, the received data vectors are rearranged so that each even-numbered sample is followed by its corresponding odd-numbered sample. A similar rearrangement is performed for the channel response matrix as shown in Equation 5a.

チップ速度のＮ倍の速度でのサンプリングについては，式５bが配列となる。

For sampling at N times the chip speed, Equation 5b is an array.

ここで、d(i)は拡散ベクトルｄのｉ番目の要素である。拡散データベクトルの長さはＮsである。拡張前進代入手法を用いると、ｄ(0)、d＾(0)を算定するためのゼロ強制解が式６aおよび式７aにより表わされる(５２)。式６ａは

Here, d (i) is the i-th element of the diffusion vector d. The length of the spread data vector is Ns. When the extended forward substitution method is used, the zero compulsory solution for calculating d (0) and d ^ (0) is expressed by Equation 6a and Equation 7a (52). Equation 6a is

ｄ(0)についての一般式である。また、式７aはd＾(0)についての零強制解である。

This is a general formula for d (0). Equation 7a is a zero compulsory solution for d ^ (0).

同様にして、チップ速度のＮ倍の速度については、式６b及び式７bを用いる。

Similarly, Expressions 6b and 7b are used for a speed N times the chip speed.

式７a及び式７bを解く際に、後での使用のため、図示のとおりv^Hを算定するが、式７aについてのv^Hは式８により算定し、蓄積する(５２)。

When solving equation 7a and Formula 7b, for later use, but to calculate the following v ^H shown, v ^H for formula 7a are determined using Equation 8, and stores (52).

d＾(0)は式９によりv^Hを用いて算定される。

d ^ (0) is determined using v ^H by Equation 9.

H行列のトプリッツ(Toeplitz)構造を用いて、残りの拡散データ要素を零強制手法で連鎖的に式１０（a）により算定することができる(５４)。

Using the Toeplitz structure of the H matrix, the remaining spread data elements can be calculated in a chain by Equation 10 (a) using the zero forcing method (54).

チップ速度のＮ倍については、式１０（b）を用いる。

For N times the chip speed, Equation 10 (b) is used.

拡散データベクトルの算定のあと、それぞれの通信信号バーストデータを、例えば拡散データベクトルとそれぞれのバースト符号との混合などにより逆拡散する(５６)。

After the calculation of the spread data vector, each communication signal burst data is despread by mixing the spread data vector and each burst code, for example (56).

  The complexity of using the extended forward substitution method is summarized in Table 1, excluding the despreading process.

Table 1
・ Calculation of v ^H ... 4 multiplications and 1 reciprocal calculation ・ Calculation of d ^ (0) ... 2 multiplications

・ Calculation of d ^ (1) ... 4 multiplications
・ Each calculation up to d ^ (W−1)… 2 multiplications ・ Calculate each d ^ (i) from d ^ (w) to d ^ (Ns−1)… (2W + 2) multiplications ・ All Number of multiplications: 2Ns + (W-1) / W + 2W .. (Ns-W + 1)
・ Total number of calculations: 2Ns + (W-1) ・ W + 2W .. (Ns−W + 1) +5

For TDD burst type 2, Ns is 1104 and W is 57. Therefore, to solve d using extended forward substitution 200 times per second, 99.9016 hundreds per second in the case of sampling at twice the chip rate. For real operations (MROPS) and sampling at chip speed, 44.95 MROPS actual operations are required.

  Another method for estimating the data is a method of approximately banding the Cholesky method, which will be described with reference to FIG. The cross-correlation matrix R is calculated to be a square (Ns × Ns), and is formed into a band according to Equation 11 (58).

ここで(・)^Hはエルミート関数を示す。Ｈは２(Ns＋W−１)×Nsの大きさである。式１１は、チップ速度の２倍の速度でサンプリングする場合は式１２（a）に書き替えられる。

Where (·) ^H denotes a Hermitian function. H is 2 (Ns + W-1) × Ns. Equation 11 is rewritten as Equation 12 (a) when sampling at twice the chip speed.

チップ速度のＮ倍の速度でのサンプリングについては、式１２（b）を用いる。

For sampling at a speed N times the chip speed, Expression 12 (b) is used.

式１２aまたは式１２bを用いて得られるＲは、大きさＮs×Ｎs型の正方行列で、かつ帯状に区分けされ、チップ速度の２倍の速度でのサンプリングについては、式１３に示すように、Ｗ=３でＮs=10である。

R obtained using Equation 12a or Equation 12b is a square matrix of size Ns × Ns, and is divided into strips. For sampling at twice the chip speed, as shown in Equation 13, W = 3 and Ns = 10.

一般にＲの帯域幅は式１４により得られる。

In general, the bandwidth of R is given by Equation 14.

近似コレスキー手法を用いるとともに、Ｒの部分ブロックＲ_subにN_col× N_colの大きさの行列を用いる。Ｒ_subの通常の大きさは(２Ｗ−１)×(２Ｗ−１)であるが、他のサイズの行列も使用できる。部分ブロック行列Ｒ_subは式１５によるコレスキー分解を用いて分解される(６０)。

With using an approximate Cholesky approach, N _col × in partial block R _sub of R Use a matrix of size N _col . The normal size of R _sub is (2W−1) × (2W−1), but matrices of other sizes can be used. The partial block matrix R _sub is decomposed using Cholesky decomposition according to Equation 15 (60).

コレスキー係数Ｇは大きさN_col× N_colである。Ｗ＝３で大きさ５×５のＧ行列は、式１６で表される。

The Cholesky coefficient G is the size N _col × N _col . A G matrix of W = 3 and a size of 5 × 5 is expressed by Expression 16.

ここで、Ｇ_ijはＧ行列のｉ番目の行、ｊ番目の列の要素を示す。Ｇ行列は、Ｇの最後の行の後に、その底部行を右側に一要素分ずつずらして加えて行くことにより、Ｎs×Ｎsの行列Ｇ_fullにまで拡張される(６２)。Ｎs＝１０の場合では、式１６は式１７の様に拡張される(６２)。

Here, G _ij represents an element of the i-th row and j-th column of the G matrix. G matrix is after the last row of G, by gradually adding shifting the bottom row one by one element minutes on the right side, is extended to the matrix G _full of Ns × Ns (62). In the case of Ns = 10, equation 16 is expanded as equation 17 (62).

拡張データベクトルを、前進代入手法及び後退代入手法により算定する(６４)。前進代入手法は、チップ速度の２倍でのサンプリングに対し、式１８（a）によりｙを算定するのに用いられ、チップ速度のＮ倍でのサンプリングに対しては式１８（ｂ）により同様にｙの算定に用いられる。

The extended data vector is calculated by the forward substitution method and the backward substitution method (64). The forward substitution method is used to calculate y according to equation 18 (a) for sampling at twice the chip speed, and similar to equation 18 (b) for sampling at N times the chip speed. Used to calculate y.

後退代入手法は、引き続き式１９により拡散データベクトルを求めるのに用いられる。

The backward substitution technique continues to be used to determine the diffuse data vector according to Equation 19.

拡散データベクトルｄを算定したあと、各バーストデータを逆拡散処理により算定する(６６)。

After calculating the spread data vector d, each burst data is calculated by despreading processing (66).

The complexity of the approximate Cholesky decomposition method is shown in Table 2 for sampling at twice the chip speed, excluding the despreading process.
Table 2
Calculation item Calculation count / H ^H H calculation… W (W + 1)
- Cholesky decomposition operation _{... N col (W-1)} 2/2 + 3
N _col (W−1) / 2− (W−1) / 3−
(W−1) ² −2 (W−1) / 3
・ Calculation of H ^H r 2NsW
-Forward substitution: [Ns- (W-1) / 2] W and Ns (reciprocal calculation)
・ Backward substitution: [Ns− (W−1) / 2] W and Ns (reciprocal calculation)

For TDD burst type 2, Ns is 1104 and W is 57. Therefore, to execute the striped approximate Cholesky method 200 times per second for sampling at twice the chip speed, 272.56 MROPS Real operations are required. In contrast, when the zonal Cholesky method is strictly executed, 906.92 MROPS is required. For sampling at the chip speed, 221.5 MROPS is required when the banded approximate Cholesky technique is used.

  Yet another method for data detection, the Toeplitz method (Levinson-Durbin type algorithm) is used, which is illustrated in FIG. Equations 12 (a) and 12 (b) are rewritten here. For N times the chip speed, Equation 12b is used.

Ｒ行列は対称で、ｐ＝Ｗ−１の帯域幅を有するトプリッツ型である(６８)。Ｒ行列の最上部の左角にある要素Ｒ(k)は、ｋ×ｋ行列で式２０に示すように算定される。

The R matrix is symmetric and is of toplit type with a bandwidth of p = W−1 (68). The element R (k) at the upper left corner of the R matrix is calculated as shown in Equation 20 using a k × k matrix.

また、もう一つのベクトルＲ_kは、Ｒ行列の要素を用いて式２１により算定される(７０)。

Another vector R _k is calculated by Equation 21 using the elements of the R matrix (70).

太字はその添字までの全ての要素を含む行列であることを示す。(k＋１)次の段階では、この系は式２２により解かれる。

Bold indicates that the matrix includes all elements up to the subscript. In the next stage (k + 1), the system is solved by Equation 22.

[H^Hr]_k＋1はH^Hrの最初の(k＋１)番目の要素である。ｄ(k＋１)は式２３に示すように、長さkのベクトル要素ｄ(k＋１)とスカラー量d₂(k＋１)とに分解される。

[H ^H r] _{k + 1} is the first (k + 1) -th element of H ^H r. As shown in Expression 23, d (k + 1) is decomposed into a vector element d (k + 1) having a length k and a scalar quantity d ₂ (k + 1).

行列Ｒ(k＋１)は式２４のように分解される。

The matrix R (k + 1) is decomposed as shown in Equation 24.

ここで、Ｅkは交換行列(exchanger matrix)である。交換行列のベクトルへの作用は、そのベクトル要素の逆置換により生ずる。

Here, Ek is an exchange matrix. The effect on the vector of the exchange matrix is caused by the inverse permutation of its vector elements.

  Using the Yule-Walker equation for linear prediction, Equation 25 is obtained.

次数再帰(recursion)手法を用いると、式２６、式２７及び式２８が得られる。

Using the order recursion technique, Equation 26, Equation 27, and Equation 28 are obtained.

ｙ(ｋ)を使用して、ｄ(ｋ＋１)が式２９、式３０及び式３１により算定する (７４)。

Using y (k), d (k + 1) is calculated by Equation 29, Equation 30, and Equation 31 (74).

ここで、(H^Hr)_k＋1はH^Hrの(k＋1)番目の要素である。

Here, (H ^H r) _{k + 1} is the (k + 1) -th element of H ^H r.

  After starting the recursive method accurately, recursive operations are sequentially performed with k = 1, 2,..., Ns. d (Ns) is the solution of Equation 32 (74).

拡散データベクトルｄをバーストのチャネルリゼーション符号で逆拡散し、データが復元する(７６)。

The spread data vector d is despread with a burst channelization code to restore the data (76).

The striped structure of the R matrix affects the recursive operation as described below. Here, R (2) and R ₂ are represented by Expression 33.

式２７及び式３０での内積計算は、各２回の乗算を必要とする。説明のためk=6の場合での式２０のＲ行列を式３４に示す。

The inner product calculation in Equation 27 and Equation 30 requires two multiplications each. For the sake of explanation, the R matrix of Equation 20 in the case of k = 6 is shown in Equation 34.

ベクトルＲ₆の０値ではない要素数は、行列の帯幅pに等しい。式２７及び式２８のそれぞれの内積、Ｒ₆ ^HＥ₆y(k)、Ｒ₆ ^HＥ₆d(k)を計算する場合、p回(k回ではない)だけ乗算すればよい。式２６及び式２９の再帰計算に関しては、演算上の効率化は全く得られない。

The number of non-zero elements of the vector R ₆ is equal to the matrix bandwidth p. When calculating the inner product, R ₆ ^H E ₆ y (k), R ₆ ^H E ₆ d (k) of each of Expression 27 and Expression 28, it is only necessary to multiply p times (not k times). With respect to the recursive calculation of Equation 26 and Equation 29, no operational efficiency is obtained.

  Table 3 shows the complexity when executing the Toplitz method.

Table 3
Calculation item Number of calculations MROPS
・ Executed once per burst
Function, calculation of H ^H H
1.3244
・ Yule-Walker equation for y
Calculation 672,888 × 100/10 ⁶ 269.1552
A function executed twice per burst, calculation of H ^H r 100.68
・ R (k + 1) d (k + 1) H ^H r calculation 672,888 × 100/10 ⁶ 538.3104

The total MROPS for the Toplits approach for the TDD burst type is 909.5456MROPS for sampling at twice the chip rate and 858.4688 MROPS for that at the chip rate.

  Another means for data detection uses Fast Fourier Transform (FFT) and will be described with reference to FIG. When sampling at the chip rate is used, the channel matrix H is square except for edge effects. A method of obtaining a data estimated value using a circulant approximation to the H matrix, the FFT of the received vector signal r, and the channel vector H is adopted.

At a sampling rate that is a multiple of the chip rate, eg, twice, the H matrix is neither square nor circular. However, the channel correlation matrix which is the matrix shown in Equation 13
R = H ^H H (84), and its submatrix is cyclic as shown by the dotted line in equation 35a. For sampling at a rate N times the chip rate, the channel correlation matrix is expressed by Equation 35b.

R行列を循環型で近似することにより、式３６,式３７及び式３８が使用できる。

Equations 36, 37, and 38 can be used by approximating the R matrix in a circular form.

ここで、(Ｒ)₁は対角行列に拡張されたR行列の第１の列である。第１の列を用いて記述されるが、この手法はR行列のどの列でも使えるように修正できる(８６)。しかし、０値ではない要素を最も多く有する列,例えばR₁、R₂、R₀、R₁、R₂と要素が並ぶ様な列を用いるのが望ましい。これらの列は、通常、少なくとも、行列の両側からW番目にある列のどれかで、例えばW番目と(Ns−W−1)番目の列を含めて、それらの間にある列である。式３８及び式３９が零強制等化手法で使用される。

Here, (R) ₁ is the first column of the R matrix expanded to a diagonal matrix. Although described using the first column, this technique can be modified to use any column of the R matrix (86). However, it is desirable to use a column having the most non-zero elements, for example, a column in which elements such as R ₁ , R ₂ , R ₀ , R ₁ , R ₂ are arranged. These columns are usually at least one of the Wth columns from both sides of the matrix, including the Wth and (Ns−W−1) th columns, for example. Equations 38 and 39 are used in the zero forced equalization technique.

Dは直交離散フーリエ変換(DFT)行列であることから、式４０、式４１、式４2が導かれる。従って、d^は式４３、式４４及び式４５（a）により、フーリエ変換を

Since D is an orthogonal discrete Fourier transform (DFT) matrix, Expressions 40, 41, and 42 are derived. Therefore, d ^ is obtained by performing Fourier transform according to Equation 43, Equation 44 and Equation 45 (a).

用いて算定できる。ここで(・)₁ は第１の列を示すが、相似の式では、R行列のどの列でも使用できる。F(・)はフーリエ変換関数を示す。F(H^Hr)は式４５（b）により、

Can be used to calculate. Here, (·) ₁ indicates the first column, but in the similar expression, any column of the R matrix can be used. F (•) represents a Fourier transform function. F (H ^H r) is expressed by Equation 45 (b)

高速フーリエ変換(FFT)を行って適切に計算される。式４５（a）の結果について逆フーリエ変換、F^ー1(・)を行うと、逆拡散ベクトルが得られる(８８)。送信されたデータは、適合する符号を用いて逆拡散することにより復元される(９０)。

It is calculated appropriately by performing a fast Fourier transform (FFT). When the result of Expression 45 (a) is subjected to inverse Fourier transform, F ⁻¹ (•), a despread vector is obtained (88). The transmitted data is recovered (90) by despreading with a matching code.

Table 4 shows the complexity of the FFT method.
Table 4
Calculation item Number of calculations MROPS
-Function calculation executed once per burst, calculation of H ^H H 1.3224
・ F ([R] ₁ ) ・ Nslog ₂ Ns 11160 × 100/10 ⁶ 4.4640
・ Function calculation executed twice per burst, calculation of H ^H r by FFT
38
・ Calculation of Equation 45 0.8832
・ F- ¹ (d) ・ Nslog ₂ Ns
8.9280

Total 55MROPS

This FFT solution is less complicated than other methods. However, a decrease in accuracy results from the cyclic approximation.
In a further approach that applies FETs to solve the data vector for sampling at a multiple of the chip rate, the samples are weighted and synthesized as described in FIG. In order to explain the sampling at twice the chip speed, r ₁ is an odd sample and r ₂ is an even sample. Each element of r1, for example, the first element r ₁ (0) is weighted and synthesized with r ₂ (0) which is a corresponding element of r ₂ as shown in Expression 46. r _eff (0) is the effectively synthesized matrix r _eff
Is a modified composite element. W1, W2

は重みである。チップ速度のN倍でのサンプリングに対しては、式４７を用いる。H₁からH_nまでのチャネル応答行列に対し、類似の重み付けを行い、

Is the weight. Equation 47 is used for sampling at N times the chip speed. Similar weighting is applied to the channel response matrix from H ₁ to H _n ,

H_eff
が生成される(９２)。その結果、式３は式４８となる。もたらされた式は、

H _eff
Is generated (92). As a result, Equation 3 becomes Equation 48. The resulting formula is

式４９によるFFTで容易に解くことができるNs×Ns型の式である(９４)。

This is an Ns × Ns type equation that can be easily solved by the FFT according to Equation 49 (94).

逆フーリエ変換を用いて拡散データベクトルが算定される。次に、バーストデータが、バースト符号を用いた拡散データベクトルの逆拡散により得られる(９６)。式４９はH_eff 行列の最初の列を用いるが、この方法はH_effの代表的な列を用いるように修正できる。
高速フーリエ変換(FFT)を用いる更に他の解法は零埋め込みを利用するもので、これは図９を用いて説明する。式５はデータベクトルにおける全ての他の要素、例えば偶数要素のようなものを０値とする零埋め込みにより修正される(９８)。修正されたｄ行列はd~で表す。Ｈ行列も拡張して、Ｈ~で示す。Ｈ行列の拡張は、各列をその列の右側に繰返し、かつ繰返された列の各要素を１行づつ下にずらし、さらにずらした列の上端に０値を埋め込むことにより行う。式４９aは、Ｗ＝３、Ｎs＝４の場合で、チップ速度の２倍の速度でのサンプリングに対してのこのような系の表示である。

A diffusion data vector is calculated using an inverse Fourier transform. Next, burst data is obtained by despreading the spread data vector using a burst code (96). Equation 49 uses the first column of the H _eff matrix, but this method can be modified to use a representative column of H _eff .
Yet another solution using Fast Fourier Transform (FFT) utilizes zero padding, which will be described with reference to FIG. Equation 5 is modified by zero padding with zero values for all other elements in the data vector, such as even elements (98). The modified d matrix is denoted by d ~. The H matrix is also expanded and denoted by H ~. The expansion of the H matrix is performed by repeating each column to the right side of the column, shifting each element of the repeated column down by one row, and further embedding a zero value at the upper end of the shifted column. Equation 49a is a representation of such a system for sampling at twice the chip rate with W = 3 and Ns = 4.

チップ速度のN倍の速度でのサンプリングに対しては、式４９（b）を用いるが、この場合,簡単のため、Ns＝３としている。

For sampling at a speed N times the chip speed, Equation 49 (b) is used. In this case, for simplicity, Ns = 3.

一般には、チップ速度のN倍に対するＨ~行列は、(ＮＮs)× (ＮＮs)型である。行列Ｈ~は正方、トプリッツ型で、かつ近似的に循環型であり、その大きさは２Ｎs×２Ｎsである。零強制解を式５０に示す(１００)。

In general, the H ~ matrix for N times the chip speed is of the (NNs) x (NNs) type. The matrix H˜ is square, toplit, and approximately circular, and its size is 2Ns × 2Ns. The zero forced solution is shown in Equation 50 (100).

第１列以外の列が類似の高速フーリエ変換(FFT)では使用できる。更に最初のチャネル応答行列Ｈの列、或いはＨの列から導出されたＨ~の推定拡張された列のいずれも使用できる。d~のＮ番目ごとの値を用いてｄを推算する。次に、適合する符号を用いて、拡散データベクトルｄを逆拡散し、データを復元する(１０２)。

Columns other than the first column can be used in similar fast Fourier transform (FFT). Furthermore, either the first column of the channel response matrix H or the estimated extended column of H ~ derived from the column of H can be used. d is estimated using every Nth value of d ~. Next, the spread data vector d is despread using a suitable code to restore the data (102).

  It can be applied to a receiver of a large capacity communication system such as a time division duplex CDMA (TDD / CDMA) communication system.

Explanatory drawing of the radio | wireless communications system which consists of the base station and user apparatus which are arrange | positioned for every working range Simplified transmitter and single user receiver Illustration of communication burst Flow diagram of extended forward substitution method for single user detection (SUD) Flow diagram of striped approximate Cholesky method for single user detection (SUD) Flow diagram of Toplit's method for single user detection (SUD) Flow diagram of Fast Fourier Transform (FFT) technique applied to channel correlation matrix for single user detection (SUD) Flow diagram of Fast Fourier Transform (FFT) technique for single user detection (SUD) using effective synthesis method Flow diagram of Fast Fourier Transform (FFT) method for single user detection (SUD) using zero embedding

Explanation of symbols

10 wireless communication system 12 base station 14 user equipment 16 communication signal burst 18 guard period 20 midamble
22, 24 Data area 26 Transmitter 28 Receiver 30 Wireless transmission channel 32 Data generator 34 Modulation / spreading device 36 Modulator 38, 40 Antenna 42 Demodulator 43 Sampling device 44 Channel estimation device 46 Single user detection (SUD) device 48 Sampling the received signal r at a rate that is a multiple of the chip rate 50 Calculation of the channel response matrix corresponding to sampling at a rate that is a multiple of the chip rate 52 First of the spread data vector using the spread forward substitution and zero forcing methods Calculate element d ^ (0). d ^ (0) in the calculation is also sought factor v ^H using the channel response matrix, use it. v ^H is accumulated 54 The remaining spread data vector elements d ^ (1), ..., d ^ (Ns-1) are calculated. For the calculation, the accumulated factor v ^H is used and the zero forcing method and the diffusion forward method are used in a chain. 56 The data is restored by despreading the diffusion data vector d. Calculate the banded matrix R 60 Execute approximate Cholesky decomposition for sub-block R _sub of R matrix 62 Expand Cholesky coefficient G and generate full matrix G _full 64 Forward substitution method and reverse Using the substitution method, the spread data vector d is calculated. 66 Data is restored by despreading from the spread data vector d. 68 Using the channel response matrix, the symmetrical and toplit R matrix is calculated. 70 Using the R matrix, Matrix R _k is calculated 72 (k + 1) In the next stage, vectors d ₁ (k + 1) and y ₁ (with length k and scalar quantities d ₂ (k + 1) and y ₂ (k), respectively. k) It allows calculating a vector d (k + 1). Further, after determining d (k + 1), the calculation 74 d (Ns) is sequentially estimated until d (Ns). The spread data vector 76 The data is restored by despreading from the spread data vector d. 84 Channel response matrix and its own Combining transpose matrix to calculate channel correlation matrix R 86 Using column with R matrix, approximate R matrix with circular form 88 With column with channel correlation matrix or approximate circular matrix and received vector The spread data vector d is calculated by Fourier transform of a column of the channel response matrix multiplied by. Inverse Fourier transform 90 for the obtained result Data is restored by despreading from the spread data vector d 92 The effective chip speed is obtained by weighting the samples at a multiple of the chip speed and the corresponding channel response matrix elements. The samples are combined and the result is an effective received signal vector and channel response matrix. 94 A representative column of effective channel response matrix and the received vector, respectively, Fourier transform of the spread data vector d Calculation. Next, the obtained result is inverse Fourier transformed 96. The data is restored by despreading from the spread data vector d. 98 0 value embedding 100 reception in the data vector so that the data vector has the same length as the received data symbol. The zero-value embedded data vector is determined by the Fourier transform of each of the generated vector and the representative column of the combined channel correlation matrix. Next, the obtained result is inverse Fourier transformed. 102 Data is restored by despreading from the spread data vector d.

Claims (69)

A method used when receiving a plurality of data signals transmitted from a transmitting side via a shared frequency spectrum in a code division multiple access communication system, each receiving a similar channel response,
Receiving a composite signal of the transmitted data signal via the shared frequency spectrum;
Sampling the composite signal at a rate that is a multiple of the chip speed of the composite signal;
Estimating a channel response to the composite signal at a rate that is a multiple of the chip rate;
Calculating a cross-correlation matrix using the estimated channel response;
Calculating a first spread data estimate using the elements from the cross-correlation matrix and recursively calculating subsequent estimates using additional elements of the cross-correlation matrix. The process of calculating using degree recursion with and
Estimating the data of the data signal using the spread data vector.   The method according to claim 1, wherein the spread data estimated value is calculated by combining a scalar part and a vector part of the spread data estimated value.   The method according to claim 1, wherein the spread data vector is calculated using a Yule-Walker equation.   The method according to claim 1, wherein the first spread data estimated value is calculated using an element at an upper left end of the cross-correlation matrix. A code division multiple access receiver used when receiving a plurality of data signals transmitted from a transmitting side via a shared frequency spectrum and each receiving a similar channel response,
Means for receiving a composite signal of the transmitted data signal via the shared frequency spectrum;
Means for sampling the composite signal at a rate that is a multiple of the chip speed of the composite signal;
Means for estimating a channel response to the composite signal at a rate that is a multiple of the chip rate;
Means for calculating a cross-correlation matrix using the estimated channel response;
Calculating a first spread data estimate using the elements from the cross-correlation matrix and recursively calculating subsequent estimates using additional elements of the cross-correlation matrix. Means for calculating using degree recursion with
Means for estimating data of the data signal using the spread data vector.   6. The receiver according to claim 5, wherein the spread data estimated value is calculated by combining a scalar part and a vector part of the spread data estimated value.   The receiver according to claim 5, wherein the spread data vector is calculated using a Yule-Walker equation.   The receiver according to claim 5, wherein the first spread data estimated value is calculated using an element at an upper left end portion of the cross-correlation matrix. A code division multiple access receiver used when receiving a plurality of data signals transmitted from a transmitting side via a shared frequency spectrum and each receiving a similar channel response,
An antenna for receiving a composite signal of the transmitted data signal via the shared frequency spectrum;
A sampling device that samples the combined signal at a rate that is a multiple of the chip rate of the transmitted data signal;
A channel estimation device for estimating a channel response to the combined signal at a speed that is a multiple of the chip speed;
A cross-correlation matrix is calculated using the channel response estimate, and a first spread data estimate is calculated using elements from the cross-correlation and subsequent using an additional element of the cross-correlation matrix A single-user detector that calculates a spread data vector using degree recursion by recursively calculating an estimate of
A code division multiple access receiver that estimates data of the data signal from the spread data vector.   The receiver according to claim 9, wherein the spread data estimated value is calculated by combining a scalar part and a vector part of the spread data estimated value.   The receiver according to claim 9, wherein the spread data vector is calculated using a Yule-Walker equation.   The receiver according to claim 9, wherein the first spread data estimated value is calculated using an element at an upper left end portion of the cross correlation matrix. A method used when receiving a plurality of data signals transmitted from a transmitting side via a shared frequency spectrum in a code division multiple access communication system, each receiving a similar channel response,
Receiving a composite signal of the transmitted data signal via the shared frequency spectrum;
Sampling the composite signal at a rate that is a multiple of the chip speed of the composite signal;
Estimating a channel response to the composite signal at a rate that is a multiple of the chip rate;
Calculating a column of a channel correlation matrix using the estimated value of the channel response;
Calculating a spread data vector using the calculated sequence, an estimate of the channel response, the received composite signal and a Fourier transform;
Estimating the data of the data signal using the spread data vector.   The method of claim 13, wherein the calculated column is a first column of the channel correlation matrix.   14. The method of claim 13, wherein a length of an impulse response of the composite signal is W, and the calculated column is at least W-1 column from an end of the channel correlation matrix.   14. The method of claim 13, wherein the calculation of the spread data vector uses a Fourier transform of a product of a Hermitian function of a channel response matrix and the received composite signal.   14. The method of claim 13, wherein the calculation of the diffusion data vector uses a Fourier transform of the calculated sequence.   The method according to claim 17, wherein the calculated value of the Fourier transform of the column is multiplied by the number of spreading chips transmitted in the data signal.   The method of claim 13, wherein the Fourier transform is a fast Fourier transform.   The method of claim 19, wherein the calculation of the spread data vector further uses an inverse fast Fourier transform. A code division multiple access receiver used when receiving a plurality of data signals transmitted from a transmitting side via a shared frequency spectrum and each receiving a similar channel response,
Means for receiving a composite signal of the transmitted data signal via the shared frequency spectrum;
Means for sampling the composite signal at a rate that is a multiple of the chip speed of the composite signal;
Means for estimating a channel response to the composite signal at a rate that is a multiple of the chip rate;
Means for calculating a column of a channel correlation matrix using the estimated value of the channel response;
Means for calculating a spread data vector using the calculated sequence, an estimate of the channel response, the received composite signal and a Fourier transform;
Means for estimating data of the data signal using the spread data vector.   The receiver of claim 21, wherein the calculated column is a first column of the channel correlation matrix.   The receiver according to claim 21, wherein a length of an impulse response of the combined signal is W, and the calculated column is at least W-1 column from an end of the channel correlation matrix.   The receiver according to claim 21, wherein the calculation of the spread data vector uses a Fourier transform of a product of a Hermitian function of a channel response matrix and the received composite signal.   The receiver according to claim 21, wherein the calculation of the spread data vector uses a Fourier transform of the calculated sequence.   26. The receiver according to claim 25, wherein the calculated value of Fourier transform of the column is multiplied by the number of spreading chips transmitted by the data signal.   The receiver according to claim 21, wherein the Fourier transform is a fast Fourier transform.   The receiver according to claim 21, wherein the calculation of the spread data vector further uses inverse fast Fourier transform. A code division multiple access receiver used when receiving a plurality of data signals transmitted from a transmitting side via a shared frequency spectrum and each receiving a similar channel response,
An antenna for receiving a composite signal of the transmitted data signal via the shared frequency spectrum;
A sampling device that samples the combined signal at a rate that is a multiple of the chip speed of the combined signal;
A channel estimation device for estimating a channel response to the combined signal at a speed that is a multiple of the chip speed;
Calculating a column of a channel correlation matrix using the estimated value of the channel response, and calculating a spread data vector using the calculated column, the estimated value of the channel response, the received synthesized signal and the Fourier transform; A user detection device,
A receiver for estimating data of the data signal from the spread data vector.   30. The receiver of claim 29, wherein the calculated column is a first column of the channel correlation matrix.   30. The receiver according to claim 29, wherein a length of an impulse response of the composite signal is W, and the calculated column is at least W-1 column from an end of the channel correlation matrix.   30. The receiver of claim 29, wherein the spread data vector calculation uses a Fourier transform of a product of a Hermitian function of a channel response matrix and the received composite signal.   30. The receiver of claim 29, wherein the spread data vector calculation uses a Fourier transform of the calculated sequence.   34. The receiver according to claim 33, wherein the calculated value of Fourier transform of the column is multiplied by the number of spreading chips transmitted by the data signal.   30. The receiver of claim 29, wherein the Fourier transform is a fast Fourier transform.   30. The receiver of claim 29, wherein the calculation of the spread data vector further uses inverse fast Fourier transform. A method used when receiving a plurality of data signals transmitted from a transmitting side via a shared frequency spectrum in a code division multiple access communication system, each receiving a similar channel response,
Receiving a composite signal of the transmitted data signal via the shared frequency spectrum;
Sampling the synthesized signal at a rate that is a multiple of the chip rate of the data signal;
Synthesizing the sampling output sampled at the multiple speed into a sampling output at the effective chip speed;
Estimating a channel response to the composite signal at a rate that is a multiple of the chip rate;
A process of synthesizing a channel response estimation value estimated at the multiple speed into a channel response at an effective chip speed;
Calculating a spread data vector using a sampling output according to the effective chip speed, a channel response according to the effective chip speed and a Fourier transform;
Estimating the data of the data signal using the spread data vector.   38. The method of claim 37, wherein the sampling output and channel response estimate with a rate that is a multiple of the chip rate is weighted prior to synthesis.   38. The method of claim 37, wherein the channel response due to the effective chip rate is an effective channel response matrix.   38. The method of claim 37, wherein the spread data vector calculation uses a column of a channel response matrix derived using the response by the effective chip rate.   38. The method of claim 37, wherein the column is the first column of the channel response matrix.   38. The method of claim 37, wherein the calculation of the spread data vector further uses an inverse Fourier transform. A code division multiple access receiver used when receiving a plurality of data signals transmitted from a transmitting side via a shared frequency spectrum and each receiving a similar channel response,
Means for receiving a composite signal of the transmitted data signal via the shared frequency spectrum;
Means for sampling the composite signal at a rate that is a multiple of the chip rate of the data signal;
Means for synthesizing the sampling output sampled at the multiple speed into a sampling output at the effective chip speed;
Means for estimating a channel response to the combined signal at a rate that is a multiple of the chip rate;
Means for synthesizing the estimated channel response value at the multiple speed into a channel response at the effective chip speed;
Means for calculating a spread data vector using a sampling output according to the effective chip speed, a channel response according to the effective chip speed and a Fourier transform;
Means for estimating data of the data signal using the spread data vector.   44. The receiver of claim 43, wherein the sampling output and channel response estimate by a rate that is a multiple of the chip rate is weighted prior to synthesis.   44. The receiver of claim 43, wherein the channel response due to the effective chip rate is an effective channel response matrix.   44. The receiver of claim 43, wherein the spread data vector calculation uses one column of a channel response matrix derived using the response by the effective chip rate.   44. The receiver of claim 43, wherein the column is the first column of the channel response matrix.   44. The receiver of claim 43, wherein the spread data vector calculation further uses inverse Fourier transform. A code division multiple access receiver used when receiving a plurality of data signals transmitted from a transmitting side via a shared frequency spectrum and each receiving a similar channel response,
An antenna for receiving a composite signal of the transmitted data signal via the shared frequency spectrum;
A sampling device for sampling the synthesized signal at a rate that is a multiple of the chip rate of the data signal;
A channel estimation device for estimating a channel response to the combined signal at a speed that is a multiple of the chip speed;
The sampling output sampled at the multiple speed is combined into a sampling output form at the effective chip speed, and the channel response estimated value estimated at the multiple speed is combined into a channel response form at the effective chip speed. A receiver for estimating data of the data signal from the spread data vector, comprising: a sampling output by speed; a channel response by the effective chip speed; and a single user detection device for calculating a spread data vector using Fourier transform.   50. The receiver of claim 49, wherein the sampling output and channel response estimate by a rate that is a multiple of the chip rate is weighted prior to synthesis.   50. The receiver of claim 49, wherein the channel response due to the effective chip rate is an effective channel response matrix.   50. The receiver of claim 49, wherein the spread data vector calculation uses one column of a channel response matrix derived using the response by the effective chip rate.   50. The receiver of claim 49, wherein the column is the first column of the channel response matrix.   50. The receiver of claim 49, wherein the spread data vector calculation further uses inverse Fourier transform. A method used when receiving a plurality of data signals transmitted from a transmitting side via a shared frequency spectrum in a code division multiple access communication system, each receiving a similar channel response,
Receiving a composite signal of the transmitted data signal via the shared frequency spectrum;
Sampling the synthesized signal at a rate that is a multiple of the chip rate of the data signal;
Estimating a channel response as a channel response matrix for the composite signal at a rate that is a multiple of the chip rate;
A modified version of a spread data vector with a magnitude corresponding to a multiple of the chip rate is calculated using one column of the channel response matrix, an estimate of the channel response matrix, the output of the sampling, and a Fourier transform The process of
Estimating the spread data vector by erasing elements of the modified version such that the spread data vector estimate has a magnitude corresponding to the chip speed.   56. The method of claim 55, wherein a speed that is a multiple of the chip speed is N times the chip speed, and the estimate of the spread data vector includes a modified version separated by N elements.   57. The method of claim 56, further comprising calculating an extended version of the channel response matrix by adding N-1 columns for each column of the channel response matrix.   56. The method of claim 55, wherein the Fourier transform is a fast Fourier transform.   59. The method of claim 58, wherein the calculation of the modified version further uses an inverse fast Fourier transform. A code division multiple access receiver used when receiving a plurality of data signals transmitted from a transmitting side via a shared frequency spectrum and each receiving a similar channel response,
Means for receiving a composite signal of the transmitted data signal via the shared frequency spectrum;
Means for sampling the composite signal at a rate that is a multiple of the chip rate of the data signal;
Means for estimating a channel response as a channel response matrix for the composite signal at a rate that is a multiple of the chip rate;
A modified version of a spread data vector with a magnitude corresponding to a multiple of the chip rate is calculated using one column of the channel response matrix, an estimate of the channel response matrix, the output of the sampling, and a Fourier transform Means to
Means for estimating the spread data vector by erasing elements of the modified version such that the spread data vector estimate has a magnitude corresponding to the chip speed.   61. The receiver of claim 60, wherein the receiver includes a modified version in which a multiple of the chip rate is N times the chip rate and the estimate of the spread data vector is separated by N elements.   62. The receiver of claim 61, further comprising calculating an extended version of the channel response matrix by adding N-1 columns for each column of the channel response matrix.   61. The receiver of claim 60, wherein the Fourier transform is a fast Fourier transform.   64. The receiver of claim 63, wherein the calculation of the modified version further uses a fast Fourier inverse transform. A code division multiple access receiver used when receiving a plurality of data signals transmitted from a transmitting side via a shared frequency spectrum and each receiving a similar channel response,
An antenna for receiving a composite signal of the transmitted data signal via the shared frequency spectrum;
A sampling device for sampling the synthesized signal at a rate that is a multiple of the chip rate of the data signal;
A channel estimation device that estimates a channel response as a channel response matrix for the combined signal at a rate that is a multiple of the chip rate;
A modified version of a spread data vector with a magnitude corresponding to a multiple of the chip rate is calculated using one column of the channel response matrix, an estimate of the channel response matrix, the output of the sampling, and a Fourier transform And a single user detection device that estimates the spread data vector by erasing elements of the modified version such that the spread data vector estimate is a magnitude corresponding to the chip speed.   64. The receiver of claim 61, wherein the receiver includes a modified version in which a multiple of the chip rate is N times the chip rate, and the spread data vector estimate is separated by N elements.   68. The receiver of claim 66, further comprising calculating an extended version of the channel response matrix by adding N-1 columns for each column of the channel response matrix.   66. The receiver of claim 65, wherein the Fourier transform is a fast Fourier transform.   66. The receiver of claim 65, wherein the calculation of the modified version further uses a fast Fourier inverse transform.

Images